Background
The vascular endothelium in the brain is the cellular interface between the blood and the central nervous system. These cells along with pericytes and vascular astrocytes form an essential physical barrier to immune cells, bacterial and viral infections (reviewed by [
1,
2]). In addition, the nutrient homeostasis of the CNS is governed by the vascular endothelium, which is selective to or permeable to a range of small molecules and drugs [
3]. Collectively, this is often referred to as the blood-brain barrier (BBB). Rather than being an impenetrable fortress, the BBB serves a range of critical functions including maintaining the brain’s immune privileged status (low frequency of leukocytes) and tightly regulating the transfer of nutrients and waste products [
4,
5].
The main cellular component of the BBB is an uninterrupted layer of microvascular endothelial cells that are joined by tight-junctions (reviewed by [
6]). These cells are highly polarised, with their basolateral aspect juxtaposed to a continuous protein-containing basement membrane (basal lamina). The presence of tight-junction complexes (e.g. claudin-5, occludin, ZO-1 (see [
7]) seals adjacent endothelial cells and restricts paracellular transfer of large molecules and cells [
6]. The apical aspect of the vascular endothelium is in direct contact with the blood and is therefore exposed to a range of insults including infections and systemic inflammatory responses.
It has become clear that the BBB is affected or changed in a wide variety of neurological diseases (e.g. multiple sclerosis [
8,
9] and HIV infection [
10,
11]), and in turn may modulate the disease process or clinical outcome [
10,
12‐
14]. Although endothelial cells are not regarded as classical immune cells, extensive literature demonstrates that endothelial cells are inextricably involved in the inflammatory responses in many tissues. This is most likely very true in the human brain; however, most of what we believe about human brain endothelial (BBB) function has been inferred from observations using non-human or non-CNS-derived endothelial cells. This includes endothelial cells from other species [
15‐
17] or from other tissues, such as the umbilical vein [
18]. Understandably, the number of studies that have actually used human brain endothelial cells (either primary or immortalised) represents a minor proportion of the literature.
Inflammatory cytokines regulate the function of brain microvascular endothelial cells leading to changes in endothelial activation [
19], inflammatory phenotype [
19] and extravasation of immune cells [
19‐
21]. Clinically, this likely contributes to the severity of disease where BBB dysfunction is evident. However, the temporal profile of these changes has not been studied in detail using human endothelial cells from the brain (or of brain origin). Here we compare the temporal response profiles of brain endothelial cells to several major pro-inflammatory regulators (IL-1β and TNFα) and reveal that the nature of the inflammatory response differs substantially. Initially, we used xCELLigence biosensor technology, which revealed that the global temporal response of the brain endothelial cells to IL-1β and TNFα was different. This observation was corroborated where expression or secretion of a number of key pro-inflammatory mediators were differentially regulated by IL-1β and TNFα. Using an extensive cytokine cytometric bead array (CBA) panel and flow-cytometry panel, we also advance our understanding of which cytokines, chemokines and adhesion molecules are expressed by brain endothelial cells. IL-1β or TNFα also induced changes in endothelial barrier strength measured as transendothelial electrical resistance (TEER) using electric cell substrate impedance-sensing (ECIS) technology. We observed complex differences in the changes in TEER induced by IL-1β and TNFα, where there were differences in the magnitude of the response and temporal nature of the response.
Discussion
To our knowledge this study represents the most comprehensive simultaneous analysis of cell-surface adhesion molecules and cytokine secretion following pro-inflammatory cytokine activation of human brain microvascular endothelial cells (hCMVECs). The goal was to ascertain the consequence of inflammatory activation on a broad range of immune mediators involved in BBB neuroinflammation, leukocyte communication and recruitment. The rationale was to advance our understanding of the repertoire of adhesion molecules, cytokines and chemokines expressed by brain endothelial cells and how these change in a temporal manner following activation by key inflammatory triggers (i.e. TNFα and IL-1β). Specific expression profiles for human brain endothelial cells is limited [
27‐
30] (see Table
3), in comparison to endothelial cells derived from other tissues (e.g. HMEC-1 skin endothelial cells).
The brain endothelial cells used in this study express all of the prototypic adhesion molecules expected of endothelial cells; including CD31 (PECAM), CD34, CD44, CD144 (VE-cadherin) and the tight-junction-associated protein ZO-1. In addition, the hCMVECs have very low levels of CD62E, ICAM-1 and VCAM-1 under basal non-inflamed conditions. This is consistent with the phenotype of brain endothelial cells in vivo [
31] and confirms that our in vitro culture conditions maintain the cells in a non-activated state. This is extremely important for being able to induce inflammation. As anticipated, both IL-1β and TNFα increased cell-surface expression of these key leukocyte-adhesion molecules; however, the potency of the response and temporal nature was different for induction of ICAM-1 and VCAM-1. Both were up-regulated to a greater extent and for longer by IL-1β. Elevated expression levels of CD62E, ICAM-1 and VCAM-1 are reliable indicators that endothelial cells have been activated in some manner. However, the temporal window of their expression is very specific. Most of the other adhesion molecules appeared unaffected by activation with IL-1β or TNFα. However, it is worthy to note that this conclusion is drawn from flow-cytometry analysis. As such, it is possible that compartmentalised localisation may change, which could then affect the function of the adhesion molecule.
Brain endothelial cells are the cellular interface between the blood and CNS tissue and like most cells with innate immune functions can release a myriad of cytokines in response to inflammatory challenge or damage. Here, we measured the release of 38 different factors, which were comprised of cytokines, chemokines, growth factors and soluble forms of receptors/adhesion molecules (see Table
1). In total, 13 of these were secreted, with most only after activation by IL-1β or TNFα. Interestingly, we observed the selective secretion of two CCL-class chemokines, MCP-1 (CCL2) and RANTES (CCL5), but no secretion of Eotaxin (CCL11), MIP-1α (CCL3) or MIP1β (CCL4). These are, however, sufficient to target a broad spectrum of immune-cell subsets expressing CCR1, CCR2, CCR3 and CCR5. Fractalkine (CX3CL1), the only known CX3CL chemokine, was not secreted by the endothelial cells under basal or activated conditions.
The CXCL chemokines (in our panel) released by the hCMVECs were IL-8 (CXCL-8) and IP-10 (CXCL10), whereas there was no secretion of either MIG (CXCL9) or I-TAC (CXCL11). IL-8 is a potent chemoattractant of neutrophils, which signals through the CXCR1 and CXCR2 receptors, IP-10 signals through CXCR3, which is expressed by subsets of T cells and monocytes/macrophages. Of particular importance to note was that the secretion of IP-10, GCSF and GMCSF, were almost exclusively induced by IL-1β, whereas the effects of TNFα were negligible in comparison. The increased GCSF and GMCSF maybe associated with creation of an environment within the CNS to support monocyte/macrophage differentiation post extravasation. The endothelial cells secrete an abundance of MCP-1 following activation, which is a potent chemokine for monocyte/macrophage subsets. In this study, TNFα increased the secretion of IL-8 and RANTES to a much greater extent than that induced by IL-1β. These differences suggest that the leukocyte recruitment signals released by the brain endothelial cells are considerably different between TNFα and IL1β responses.
Truncated versions of certain receptors and adhesion molecules can be liberated (cleaved or secreted) from the cell surface and consequently detected in the conditioned media. Here we measured the soluble forms of the IL-1 receptors, TNF receptors and ICAM-1 and VCAM-1. Soluble IL-1 receptors were not detected under any conditions; however, soluble TNFα receptors (both TNFRI and TNFRII) were detected. Interestingly, the amount was greater following IL-1β treatment, rather than TNFα treatment. The relevance of this novel finding is unclear but may relate to the desensitisation of the inflammatory environment to restrict TNFα signaling.
It is very interesting to note that TNFα appears to induce a much greater level of ICAM-1 and VCAM-1 on the surface of the endothelial cells, in comparison to IL-1β. This could be due to the fact that the level of soluble ICAM-1 and VCAM-1 present in the culture media was greater following IL-1β activation. This demonstrates that more ICAM-1 and VCAM-1 are cleaved during the IL-1β response (compare the data in Figs.
3 and
6). These adhesion molecules are the interacting partners of CD11 family members (e.g. LFA-1) and CD49d (VLA-4), respectively, which are broadly expressed by a range of leukocyte subsets. This data series nicely provides the basis to conduct adhesion/diapedesis experiments, which will be the subject of a future study. At this stage, it is exciting to hypothesise that TNFα and IL-1β differentially regulate tethering/attachment of different leukocyte subsets to the apical surface of the brain endothelium.
As well as being a comprehensive investigation of adhesion molecules important in the wider context of endothelial biology, we reveal expression of a number of adhesion molecules by hCMVEC not described previously for human brain endothelial cells. These include CD138, CD141, CD147, CD325 and CD50. This advances our understanding of their phenotype and adhesion-molecule expression profile, all of which are molecules important in endothelial functional biology.
A very important area clinically is the influence of neuroinflammatory events on vascular barrier function. Dysfunction or breakdown of the brain vascular integrity is a well-documented event in diseases such as relapsing remitting multiple sclerosis or in ischemic stroke patients, which can lead to vascular haemorrhage in some stroke patients. Surprisingly, there is a lack of studies comparing the effect of IL-1β and TNFα on barrier function using human brain endothelial cells (primary or cell line-derived) and advanced autonomous technologies such as ECIS. We therefore used ECIS-Ztheta system to monitor the effect of TNFα and IL1β on barrier function in a label-free temporal manner. Of specific importance for us was to determine whether there were any differences between the cytokine responses and to monitor the changes in barrier resistance either acutely (first few hours—24 h) or chronically over days. There was a small acute reduction in the endothelial barrier resistance in the first few hours with higher concentrations of TNFα (above 500 pg/mL) and IL-1β. This response was small, reflecting approximately 10–15 % difference in norm resistance. Reduced TEER over this period has been reported by others [
32,
33]. After this acute period, the barrier resistance of the HCMVECs increased in a cytokine, concentration and time-dependant manner. This strengthening of the barrier occurs over a period, which is (surprisingly) omitted in other ECIS studies [
32,
33]. The magnitude of the increase in barrier resistance is greater and occurs for longer for TNFα. This perhaps suggests that the brain endothelial cells are innately programmed to recognise elevated TNFα as a more dangerous signal than IL-1β and respond accordingly in terms of strengthening its integrity. This may also be the innate response to protect against any danger signals that may lead to neutrophil activation and invasion, which in terms of the brain vasculature, can be devastating. Our barrier resistance (ECIS) data for the hCMVECs highlight the importance of conducting acute and longer term analysis of barrier function, which we have achieved in a real-time autonomous manner using ECIS ZΘ. It is clear that this is an area which requires considerably more research to be conducted to understand the pathways leading to barrier protection.
This current study highlights the fact that these two potent inflammatory mediators induce different inflammatory responses from brain endothelial cells. This was originally suggested from the xCELLigence data, which measures the net changes in cellular focal adhesion. This therefore indicates that the brain endothelial cells are innately programmed to respond to different insults in specific ways. Therefore in vivo, certain diseases, infections or tissue injury may cause elevated levels of IL-1β, which may trigger different pathways and cytokine networks in the brain endothelium in comparison to conditions where TNFα is predominantly elevated (e.g. bacterial infection and autoimmune diseases).
Future considerations
The data obtained in this study will be invaluable for investigating the dynamics of leukocyte attachment under different inflamed conditions and whether IL-1β induces recruitment of different subsets of immune cells in comparison to TNFα. The xCELLigence technology provides a cutting-edge tool to investigate the real-time temporal effects of activation of brain endothelial cells. In addition, the temporal sensitivity of ECIS for investigating cytokine changes in barrier function is incredibly powerful. This is clearly a complex area in the field of neuroinflammation, where there is still much to be discovered in terms of the endothelial responses and roles in disease. This will be applied to a range of activation signals and for conducting poly-pharmacological (multiple cytokines) analysis of how brain endothelial cells are regulated under more complex inflamed conditions.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
DTK, VN, OOR, RW, RJ, SO and ESG conducted the research. DTK, RW, SO, CEA and ESG developed and contributed to the experimental design. SO, CEA and ESG contributed resources and materials. SO, CEA and ESG obtained the funding. DTK, VN, RW, RJ, CEA, SO and ESG interpreted the data. DTK, VN, RW, CEA, SO and SG prepared the figures. DTK, VN, OOR, RJ, SO, CEA and ESG wrote and edited manuscript. All authors read and approved the final manuscript.
Simon J O’Carroll and Dan Ting Kho share first authorship.